Association of albumin and cholesterol levels with incidence of hypoglycaemia in people admitted to general internal medicine units

• Published in: Diabetic medicine Vol. 35; no. 12; pp. 1735 - 1741
• Main Authors: Leibovitz, E., Wainstein, J., Boaz, M.
• Format: Journal Article
• Language: English
• Published: England Wiley Subscription Services, Inc 01.12.2018
• Subjects: Aged; Aged, 80 and over; Albumin; Blood Glucose; Cholesterol; Cholesterol - blood; Cross-Sectional Studies; Diabetes; Diabetes mellitus; Diabetes Mellitus - blood; Diabetes Mellitus - drug therapy; Diabetes Mellitus - epidemiology; Electronic medical records; Female; General Practice - statistics & numerical data; Glucose; Hemoglobin; Hospital Units - statistics & numerical data; Humans; Hypocholesterolemia; Hypoglycemia; Hypoglycemia - blood; Hypoglycemia - epidemiology; Incidence; Internal medicine; Internal Medicine - statistics & numerical data; Israel - epidemiology; Male; Medicine; Middle Aged; Patient Admission - statistics & numerical data; Risk Factors; Serum Albumin - analysis; Serum Albumin - metabolism; Israel
• ISSN: 0742-3071; 1464-5491
• DOI: 10.1111/dme.13792

Aim To study the association between serum albumin and cholesterol levels at hospital admission and incident hypoglycaemia among people admitted to internal medicine units. Methods In this cross‐sectional analysis, we examined the electronic medical records of people hospitalized in internal medicine departments. Data extracted included all glucose measurements as well as serum albumin and cholesterol, which were measured upon admission. A hypoglycaemia event was recorded for a given person if at least one glucose measurement of ≤ 3.9 mmol/l was recorded during the hospital stay. Regression analysis was used to determine which clinical measures predict hypoglycaemia. Results During the acquisition period, 45 224 people (mean age 68.9 ± 17.8 years, 49.4% male, 21.1% diabetes mellitus) were discharged from internal medicine units. Hypoglycaemia was documented in 7.5% of these individuals (15.4% of people with diabetes vs. 5.5% of those without; P < 0.001). Logistic regression showed that both serum albumin [odds ratio (OR) 0.908, 95% confidence interval (CI) 0.896–0.919; P < 0.001) and cholesterol (OR 0.938, 95% CI 0.896–0.981; P = 0.005] were significantly associated with incident hypoglycaemia. Results remained significant even after controlling for age, sex, average glucose during hospitalization, length of hospital stay, acute infection upon admission, diabetes status, haemoglobin, white blood cell count and C‐reactive levels. A combination of hypoalbuminaemia (< 35 g/l) and hypocholesterolaemia (< 3.37 mmol/l) upon admission greatly increased the risk of incident hypoglycaemia (OR 2.544, 95% CI 2.096–3.088; P < 0.001). Conclusion Hypoalbuminaemia and hypocholesterolaemia predict incident hypoglycaemia in the hospital setting among people with and without diabetes mellitus. What's new? Among people with diabetes, hypoglycaemia is attributed to over‐treatment, whereas among people without diabetes, hypoglycaemia is attributed to disease severity. Our data suggest that serum albumin and cholesterol levels, measured upon admission to internal medicine units, are predictors of hypoglycaemia incidence and severity during hospitalization in people with and without diabetes mellitus. We argue that albumin < 35 g/l and cholesterol < 3.37 mmol/l, both surrogate markers of nutrition, strongly predict hypoglycaemia, even after controlling for age, sex, creatinine, C‐reactive protein, white blood cell count, haemoglobin, diabetes status, length of hospital stay, average glucose during hospitalization, and infection as reason for admission.